MULTIPLE REGRESSION PREDICTING LUNG CANCER BASED ON RISK FACTORS – A CASE STUDY FOR THE INDUSTRY
DUMITRU MIHAELA 1, PRAISLER MIRELA 1, 2, REBEGEA LAURA 1, FIRESCU DOREL 3 1. "Sf. Ap. Andrei" Emergency Clinical Hospital, Galati, Department of Radiotherapy, Brailei St.177, 800830 Galati, Romania
2 "Dunarea de Jos" University of Galati, Department of Chemistry, Physics and Environment, Domneasca St. 47, 800008 Galati, Romania
3. "Sf. Ap. Andrei" Emergency Clinical Hospital, Galati, Surgery Clinic II, Brailei St. 177, 800830 Galati Romania
The objective of this study is to determine a multivariate model able to perform the identification of the tumor stage and of the histopathological type for lung cancer patients, based on predictive environmental and behavioral factors. The database was built by using clinical and personal information about 106 patients with stage III or IV lung cancer and a mean age of 58 years, who were subjected to radiotherapy at the Radiotherapy Department of „Sf. Ap. Andrei” Emergency Clinical Hospital between January and December 2010. The following factors were taken into consideration: work conditions, smoking habits and duration of the exposure to risk factors. Most of the patients (43.33%) have worked in metallurgy for a mean of 27.54 years and 86.67% have smoked for a mean period of 26.92 years. The duration of smoking and the habit of smoking itself were identified by multiple linear regression as the most important predictive factors for tumor stage and for the histopathological type, respectively. Artificial intelligence techniques were used and the results have indicated that tobacco use and the environmental factors related to the work place (e.g. metallurgical industry) are the best predictive risk factors for the incidence of lung cancer.
